Surprisal Metrics for Quantifying Perturbed Conformational Dynamics in Markov State Models

Area of Science:

• Computational Biology
• Biophysics
• Molecular Dynamics

Background:

• Markov state models (MSMs) are widely used to analyze biomolecular conformational dynamics.
• Assessing changes in transition rates due to perturbations (e.g., mutations) is crucial but challenging.
• Existing methods may not adequately distinguish changes in transition rates from shifts in state definitions.

Purpose of the Study:

• To introduce a novel surprisal metric for quantifying differences in transition rates between two related Markov state models.
• To provide a method for assessing the impact of perturbations on biomolecular dynamics independent of state definition changes.
• To demonstrate the utility of this metric in identifying affected conformational states.

Main Methods:

• Development of a surprisal metric based on relative entropy and Jensen-Shannon divergence.
• Calculation of transition counts and statistical uncertainties for MSMs.
• Application of the metric to diverse biomolecular systems, including lattice models and all-atom simulations.
• Exploration of surprisal-based adaptive sampling for uncertainty reduction.

Main Results:

• The surprisal metric effectively quantifies differences in transition rates between closely related MSMs.
• The metric successfully identifies conformational states most sensitive to perturbations in protein hairpin and peptide models.
• Surprisal-based adaptive sampling significantly reduces statistical uncertainty in divergence calculations.

Conclusions:

• The surprisal metric offers a robust way to analyze changes in biomolecular dynamics under perturbation.
• This approach provides insights into how mutations and environmental factors alter molecular kinetics.
• Surprisal-based adaptive sampling can enhance the efficiency of molecular simulation and aid in ab initio design.